Associations Between Amyloid and Tau Pathology,and Connectome Alterations,in Alzheimer’s Disease and Mild Cognitive Impairment

Abstract

Background:

The roles of amyloid-β and tau in the degenerative process of Alzheimer’s disease (AD) remain uncertain. [¹⁸F]AV-45 and [¹⁸F]AV-1451 PET quantify amyloid-β and tau pathology, respectively, while diffusion tractography enables detection of their microstructural consequences.

Objective:

Examine the impact of amyloid-β and tau pathology on the structural connectome and cognition, in mild cognitive impairment (MCI) and AD.

Methods:

Combined [¹⁸F]AV-45 and [¹⁸F]AV-1451 PET, diffusion tractography, and cognitive assessment in 28 controls, 32 MCI, and 26 AD patients.

Results:

Hippocampal connectivity was reduced to the thalami, right lateral orbitofrontal, and right amygdala in MCI; alongside the insula, posterior cingulate, right entorhinal, and numerous cortical regions in AD (all p < 0.05). Hippocampal strength inversely correlated with [¹⁸F]AV-1451 SUVr in MCI (r = –0.55, p = 0.049) and AD (r = –0.57, p = 0.046), while reductions in hippocampal connectivity to ipsilateral brain regions correlated with increased [18F]AV-45 SUVr in those same regions in MCI (r = –0.33, p = 0.003) and AD (r = –0.31, p = 0.006). Cognitive scores correlated with connectivity of the right temporal pole in MCI (r = –0.60, p = 0.035) and left hippocampus in AD (r = 0.69, p = 0.024). Clinical Dementia Rating Scale scores correlated with [¹⁸F]AV-1451 SUVr in multiple areas reflecting Braak stages I-IV, including the right (r = 0.65, p = 0.004) entorhinal cortex in MCI; and Braak stages III-VI, including the right (r = 0.062, p = 0.009) parahippocampal gyrus in AD.

Conclusion:

Reductions in hippocampal connectivity predominate in the AD connectome, correlating with hippocampal tau in MCI and AD, and with amyloid-β in the target regions of those connections. Cognitive scores correlate with microstructural changes and reflect the accumulation of tau pathology.

Keywords

Alzheimer’s disease amyloid connectome diffusion tensor imaging magnetic resonance imaging mild cognitive impairment positron emission tomography tau protein

INTRODUCTION

The accumulation of amyloid-β plaques and neurofibrillary tangles of hyperphosphorylated tau are the pathological hallmarks of AD; however, the exact role of these proteins in the neurodegenerative process remains unclear [1, 2]. The amyloid hypothesis proposes that amyloid-β deposition, present over 20 years before symptom onset, is the primary driver for AD [2 –4]. The hypothesis is supported by the early and progressive appearance of amyloid-β plaques, the pathogenic role of amyloid-β in multiple other human diseases, and genetic variants of AD characterized by increased amyloid-β production [5 –12]. However, the spatial distribution of amyloid-β pathology is different from that of neurodegeneration in AD, while the extent of amyloid-β deposition poorly predicts clinical disease, and drug therapies targeting amyloid-β have been largely ineffective in human trials [2 , 13–16]. These inconsistencies have cast doubt on the amyloid hypothesis [16].

Tau pathology, also present over 10 years prior to the clinical onset of AD, arises in the transentorhinal cortex before spreading to limbic and subsequently neocortical areas as AD progresses [1, 17]. This progression, elegantly described by Braak and colleagues (2006), corresponds with areas of degeneration and with clinical disease stage while in vitro studies demonstrate neuronal dysfunction and degeneration due to tau toxicity [1 , 18–21].

Positron emission tomography (PET) imaging allows the in vivo visualization of amyloid-β and tau pathology. [¹⁸F]AV-45 (Florbetapir), [¹¹C]PiB (C-Pittsburgh compound B), and [¹⁸F]AV-1 (Florbetaben) have all been validated for visualization of amyloid-β deposition while the development of [¹⁸F]AV-1451 (Flortaucipir) has enabled the visualization of the paired helical filaments formed from hyperphosphorylated tau in AD [3 , 22–25]. These tracers are sensitive enough to demonstrate tau and amyloid-β deposition in preclinical AD while their distribution is consistent with histological studies [26 –33].

Mild cognitive impairment (MCI), an interim state between that of normal aging and dementia, provides an opportunity to examine the structural and molecular changes which occur prior to the development of AD. Patient’s with MCI have an annual conversion rate to AD of 5–10%. However, MCI represents a heterogeneous clinical group; most patients do not progress to AD after 10 years follow up, while many MCI patients will develop other forms of dementia or recover entirely [34]. Despite these limitations, MCI patients provide an enriched population, in whom a significant proportion will harbor the pathological and structural underpinnings of impending AD. Identifying these changes will not only allow us to better understand the pathological process, but also to identify those who can benefit most from novel therapies.

Diffusion weighted imaging (DWI) allows microstructural changes underlying AD and MCI to be examined in vivo. Diffusion tractography utilizes DWI to delineate white matter pathways, allowing alterations in structural connectivity between brain areas to be detected. Decreased structural connectivity has been demonstrated in the temporal lobes of patients with pre-clinical AD, with further involvement of the parietal and occipital lobes in MCI, and subsequently throughout the brain in AD [35 –39], while alterations in network topology are detectable up to 13 years before symptom onset in those with familial AD. The use of graph theory measures to examine network topology is emerging as a sensitive marker of neurodegeneration; however, the biological significance of these early topological changes remains uncertain [40 –44].

Combining these imaging modalities enables detailed examination of the pathological changes which underly AD. A number of studies have demonstrated associations between tau PET changes and DWI white matter abnormalities [45 –48], while microstructural changes within hippocampal connections have been shown to predict downstream tau accumulation only in amyloid positive individuals [49]. However, these studies often focus on specific tracts and regions, which may miss larger scale or unexpected effects, while the demonstrated effects of amyloid-β upon the neurodegenerative process have so far been unrelated to its regional distribution within the brain [38 , 51].

By combining [¹⁸F]AV-1451 and [¹⁸F]AV-45 PET with DTI in healthy controls, MCI, and AD, we aimed to examine the microstructural consequences of amyloid-β and tau pathology, and assess those changes which precede the clinical onset of AD. Examination of microstructural and cognitive changes in relation to the distribution of tau and amyloid-β pathology, both by region and according to the Braak and Thal pathological staging systems, allows scrutiny of different phases in a sequential disease process.

MATERIALS AND METHODS

Data used in the preparation of this article were obtained from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database (http://adni.loni.usc.edu) accessed in June 2019. The ADNI was launched in 2003 as a public-private partnership, led by Principal Investigator Michael W. Weiner, MD. The primary goal of ADNI has been to test whether serial MRI, PET, other biological markers, and clinical and neuropsychological assessment can be combined to measure the progression of MCI and early AD. The detailed methodology for all clinical and imaging data acquisition are available at the ADNI website (http://adni.loni.usc.edu/methods/).

Participants

A total of 86 subjects from the ADNI database were included in the study (Table 1), comprising all subjects with sufficient data to perform the below analyses as of June 2019. Cognitively normal controls (n = 28) alongside subjects with a diagnosis of MCI (n = 32) or AD (n = 26) were identified. All subjects had both [¹⁸F]AV-1451 PET, structural T₁-weighted MRI and DWI data acquisition, as specified below. A total of 78 subjects, 27 cognitive normal controls, 32 MCI and 19 AD patients, also underwent [¹⁸F]AV-45 PET imaging. Groups were matched for age and gender.

Table 1

Subject clinical characteristics

	HC		MCI		AD		HC mean versus MCI (p)	HC mean versus AD (p)	AD mean versus MCI (p)
N	28		32		26
Age, mean years (SD)	75.7	(4.9)	77.2	(7.7)	76.5	(8.4)	1.0	1.0	1.0
Gender, female (%)	20	(71)	12	(38)	9	(35)	0.009	0.007	0.82
APOE4 carrier, 1 + copies (%)^a	8	(30)	12	(40)	9	(50)	0.36	0.096	0.39
Amyloid +ve (%)^b	7	(25)	13	(41)	20	(87)	0.20	< 0.001	0.001
CDR-SB, mean (SD)	0.1	(0.2)	1.5	(1.3)	5.4	(2.6)	0.004	< 0.001	< 0.001
FAQ, mean (SD)^c	0.2	(0.8)	3.4	(3.7)	15.4	(6.7)	0.016	< 0.001	< 0.001
MMSE, mean (SD)	29.3	(1.1)	28.0	(1.8)	21.0	(4.5)	0.24	< 0.001	< 0.001
MoCA, mean (SD)	27.9	(2.1)	22.7	(2.8)	15.2	(5.0)	< 0.001	< 0.001	< 0.001
RAVLT-immediate, mean (SD)	49.5	(8.5)	31.8	(9.3)	21.0	(5.9)	< 0.001	< 0.001	< 0.001
RAVLT-forgetting, mean (SD)	4.3	(2.4)	5.1	(2.2)	4.5	(1.3)	0.30	1.0	0.74

AD, Alzheimer’s disease; CDR-SB, Clinical Dementia Rating Sum of Boxes; FAQ, Functional Activities Questionnaire; HC, healthy controls; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; MoCA, Montreal Cognitive Assessment; RAVLT, Rey Auditory Verbal Learning Test. Between group differences assessed with ANOVA, with Bonferroni post hoc correction. ^aOne HC, two MCI, and eight AD patients had unknown APOE status. ^bThree patients with AD had no amyloid-β data available. ^cOne patient with AD had no FAQ score available.

Clinical assessments

All participants completed a battery of cognitive assessments (Table 1). The Mini-Mental Status Examination (MMSE) and Montreal Cognitive Assessment (MoCA) are both 30-point assessments of global cognitive status, with low scores indicating increased impairment. The Clinical Dementia Rating Sum of Boxes (CDR-SB), another global assessment of cognition, is a variant of the Clinical Dementia Rating (CDR) scale which provides an increased range of values compared to the CDR, allowing the CDR-SB to discriminate both between and within groups of patients with AD and MCI [52]. The CDR-SB is scored out of 18, with higher scores indicating increased impairment. The Rey Auditory Verbal Learning Test (RAVLT) is a validated assessment of episodic verbal memory, capable of differentiating between and predicting the conversion from MCI to AD [53, 54]. The RAVLT-immediate assesses immediate recall of 15 words across five consecutive trials, with a maximum score of 75 indicating perfect immediate recall. The RAVLT-forgetting score examines delayed recall; scoring for those words recalled in a fifth trial, minus those recalled in a further trial, 30 minutes later; a higher score indicates more words forgotten [55]. The Functional Activities Questionnaire (FAQ) assesses a participant’s ability to perform a number of activities, such as assembling tax records, or shopping alone. The FAQ is able to sensitively distinguish patients with mild AD from those with MCI or healthy controls; higher scores indicate increased deficits [56, 57].

Blood samples were taken to assess apolipoprotein E (APOE) genotype status. The ɛ4 polymorphism of the APOE gene is the strongest genetic risk factor for late onset AD [58].

Scanning procedures

All T₁-weighted MRI, DWI MRI, [¹⁸F]AV-1451 PET, and [¹⁸F]AV-45 PET scans for each patient were collected within a mean 46-day (range 1–264 days) time window for all cognitive assessments and imaging scans.

Magnetic resonance imaging

All participants were scanned according to standardized protocols to maximize consistency across ADNI sites. 3T field strength T₁-weighted three dimensional images were acquired at 1 mm³ resolution on Siemens or General Electric (GE) MRI scanners (Siemens: magnetization-prepared rapid acquisition gradient-echo, time repetition, 2300 ms; time echo, 2.98 ms; flip angle, 9°; time inversion, 900 ms; matrix: 208×240×256 mm. GE: fast spoiled gradient-echo, time repetition, 7.2–7.4 ms; time echo, 2.94–3.02 ms; flip angle, 11°; time inversion, 400 ms; matrix: 196×256×256 mm) [59].

DWI scans were acquired at 2 mm³ resolution along 48 isotropically distributed gradient directions at b = 1000 s/mm² (Siemens: time repetition, 7200–9600 ms; time echo, 56–82 ms; flip angle, 90°; matrix: 232×232×160 mm. GE: time repetition, 7800 ms; time echo, 55–60 ms; flip angle, 90°; matrix: 232×232×138 mm). Non-diffusion-weighted (b0) images were also acquired during each scan.

Positron emission tomography

AV-1451 and [¹⁸F]AV-45 PET acquisition was performed using a number of different scanners by Siemens, GE, and Philips, as described previously [60]. Participants underwent a 20 min dynamic scan consisting of four 5 min frames acquired 50 min after administration of [¹⁸F]AV-45 at a target dose of 370 MBq±10%; and a 30 min dynamic scan consisting of six 5 min frames acquired 75 min post injection of [¹⁸F]AV-1451 at a target dose of 370 MBq±10%.

To improve consistency between different PET scanners, pre-processing of [¹⁸F]AV-1451 and [¹⁸F]AV-45 PET data was conducted within ADNI using a standardized pipeline to produce normalized images with a uniform 1.5 mm³ voxel size and isotropic resolution of 8 mm full width at half maximum [61].

Imaging data analysis

Structural MRI analysis and regions of interest

T₁-weighted images were co-registered to each subject’s diffusion weighted image space by creating a transformation matrix using FMRIB’s Linear Image Registration Tool within the FMRIB Software Library (FSL; version 5.0.8; http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/). The transformation matrix was subsequently applied by MRtrix’s mrtransform tool [62]. This method maintains T₁ resolution despite the inferior resolution of the target diffusion weighted image [62 –64]. T₁-weighted images, in diffusion space, were segmented using the Desikan-Killiany atlas and subcortical segmentation in FreeSurfer image analysis suite (version 5.3; http://surfer.nmr.mgh.harvard.edu) to produce 84 subject specific regions of interest (ROIs) [65]. Segmented data sets were visually inspected to ensure accuracy of skull stripping, segmentation, and cortical surface reconstruction; manual correction of the FreeSurfer derived pial surface and segmentation was performed where required.

The distribution of tau and amyloid-β, assessed using [¹⁸F]AV-1451 and [¹⁸F]AV-45 PET, respectively, were examined in relation to the histological staging; for tau described by Braak and colleagues [1 , 67], and amyloid-β as described by Thal and colleagues [68, 69]. The FreeSurfer derived ROIs do not provide an exact replication of the anatomical delineation of the Braak stages. The tau staging used herein therefore uses an approximation of the Braak stages derived from FreeSurfer ROIs, as previously demonstrated by Schöll and colleagues [33].

PET data analysis

Partial volume correction of [¹⁸F]AV-1451 data was performed using the Rousset Geometric Transfer Matrix method [70, 71]. This reduces the effects from the off-target binding of [¹⁸F]AV-1451 by correcting for the spillover of PET signal from areas with off-target binding, such as the choroid plexus, into neighboring areas including the hippocampus [72]. [¹⁸F]AV-1451 standardized uptake value ratio (SUVr) for each FreeSurfer derived ROI was calculated using the inferior cerebellum, derived from the SUIT template (http://www.diedrichsenlab.org/imaging/suit.htm) in Statistical Parametric Mapping (SPM; version 12; http://www.fil.ion.ucl.ac.uk/spm/), as the reference region [73, 74]. [¹⁸F]AV-45 SUVr were calculated for each FreeSurfer derived ROI with the Analyze medical imaging software package (version 12.0; Mayo Foundation AnalyzeDirect, United States) using the entire cerebellum as a reference region [75].

In light of reported off-target binding of [¹⁸F]AV-1451, the basal ganglia and thalamus were excluded from all [¹⁸F]AV-1451 analyses [76]. Additionally, the partial volume correction procedure utilized herein produces composite ROIs from the caudal and rostral middle frontal gyri; the orbitofrontal gyri and frontal pole; and the pars constituents of the inferior frontal gyrus [71]. This leaves 64 ROIs for the analyses of [¹⁸F]AV-1451 PET data, as opposed to 82 ROIs for all other analyses. To assess the extent of effect of off target binding from the choroid plexus on the hippocampus, and the adequacy of the partial volume correction on correcting this, we calculated the correlation between [¹⁸F]AV-1451 SUVr in the hippocampi and the adjacent choroid plexus, before after partial volume correction.

Separately, to determine amyloid-β positivity, we used [¹⁸F]AV-45 SUVr values calculated within ADNI using a composite cortical ROI and using cut off value derivation as previously described [77 –79]. [¹⁸F]AV-45 SUVr > 1.11 was defined amyloid positive. Four AD patients, and one control subject, without [¹⁸F]AV-45 SUVr data available ([¹⁸F]AV-1 was used instead), with those with a [¹⁸F]AV-1 SUVr > 1.08 designated amyloid positive.

Connectome construction

Pre-processing of DWI was performed using MRtrix and FSL Diffusion Toolbox to correct for noise, eddy current distortion, subject motion, and B1 field inhomogeneity [80 –83]. Probabilistic anatomically constrained tractography was performed using constrained spherical deconvolution in MRtrix [62 , 85]. Briefly, response functions for spherical deconvolution were calculated prior to estimation of the fiber orientation distributions. Tractography was performed utilizing the iFOD2 probabilistic algorithm with the subject specific five-tissue-type segmentation produced from subject specific FreeSurfer derived tissue segmentations [44]. Seed points were determined dynamically utilizing the spherical-deconvolution informed filtering of tractograms (SIFT) model until ten million completed streamlines were generated [43 , 85]. The SIFT2 algorithm was then used on all tractograms, this improves the anatomical validity of whole brain tractograms and allows streamline count to be used as a marker of connection density [43, 86]. Finally, connectome matrices were created by assigning each streamline termination to the nearest FreeSurfer defined ROI (within a 2 mm radius) and calculating the number of streamlines connecting each pair of regions (edge weight). Connections to ipsilateral and contralateral regions are included in all analyses, unless specified. Connectome visualization was performed with Circos [87], and all connectome measures were derived with the Brain Connectivity Toolbox [88].

To detect more subtle changes in connectivity, a further analysis examined alterations in connectivity between the hippocampus and ipsilateral limbic structures implicated in emotion and cognition [89, 90]. Ipsilateral connections between the hippocampi and accumbens area, amygdala, cingulate cortex, entorhinal cortex, frontal pole, insula cortex, orbitofrontal cortex, parahippocampal gyrus, and thalamus were examined. These regions encompass a broad selection of regions implicated in the limbic processing, excluding the mamillary bodies as they are not included in the FreeSurfer segmentation [91, 92].

The strength, also referred to as weighted degree, of a region is the sum weight of all connections to the region [88]. This is calculated by counting the total number of connections from each region to all other regions.

Regional and global correlations

To assess whether regions with higher tau or amyloid-β burden have alterations in connectivity, the correlation between [¹⁸F]AV-1451 SUVr, [¹⁸F]AV-45 SUVr, and strength was examined. Correlations were examined both regionally and globally. Regional correlations examine the correlation within each ROI separately. In light of the symmetrical distribution of amyloid-β and tau, and to enable detection of correlations between [¹⁸F]AV-1451, [¹⁸F]AV-45, and strength within each region while reducing the number of multiple comparisons, bilateral regions were examined. Global correlations examine the correlation between [¹⁸F]AV-1451 SUVr, [¹⁸F]AV-45 SUVr, and strength across all brain regions together at the same time, allowing any global effect of these proteins on strength or on each other to be examined. Prior to examination of global correlations, the strength of each ROI, for each subject, was normalized as a percentage of the control mean of that region. Without this correction larger ROIs, in general, would have greater strength. This normalized value is also used to aid visualization of strength values within all figures.

To determine if global amyloid-β deposition is related to hippocampal tau deposition, or hippocampal connectivity, we examined the correlation between both hippocampal [¹⁸F]AV-1451 SUVr, and hippocampal strength, with [¹⁸F]AV-45 SUVr in a composite region, comprising the frontal, cingulate, parietal, and temporal regions bilaterally. This composite region was selected due to the relative susceptibility of the contained regions to amyloid-β deposition [77, 93].

Statistical analysis

Statistical analysis and graph illustration were performed with SPSS (version 20 Chicago, IL, USA) and GraphPad Prism (version 6.0c) for MAC OS X, respectively. Connectomes with 82 ROIs (nodes) have 3,321 possible edges. Therefore, standard statistical tests are underpowered when correcting for multiple comparisons at both the edge and ROI level. Controlling the False Discovery Rate (FDR) using the Benjamini and Hochberg equation allows for correction of multiple comparisons while identifying significant between-group differences at individual edges and ROIs [94 –96].

Connectomes were thresholded by removing the weakest 0.5%of all edges by edge weight; this minimizes noise and removes clearly unconnected regions from the analysis. Between-group differences in thresholded connectomes at the edge level, connectome metrics, and [¹⁸F]AV-1451 and [¹⁸F]AV-45 PET SUVr at the nodal level were calculated using analysis of covariance (ANCOVA) including age and gender as covariates in all analyses.

Covariate-adjusted Pearson product-moment correlation analysis was used to determine the relationship between connectome measures, [¹⁸F]AV-1451 and [¹⁸F]AV-45 PET SUVr, and cognitive scores at the nodal level, while controlling for age and gender. To ensure that correlations were not driven by outlier variables, outlying values were identified using robust regression and outlier removal (ROUT, Q = 1%) in GraphPad Prism [97]. The linear relationship between these variables was confirmed by visual assessment of the scatter plots. The correlation between hippocampal connectivity and [¹⁸F]AV-45 SUVr in the target region of those hippocampal connections was assessed with Spearman’s rank order correlation, this was used due to a non-linear relationship between these variables.

To examine if APOE ɛ4 allele carrier status effects the regional distribution of tau and amyloid-β deposition, or connectome alterations, its effect on regional [¹⁸F]AV-1451 SUVr, [¹⁸F]AV-45 SUVr, and strength were examined using ANCOVA, with age and gender as covariates in all analyses.

FDR correction for multiple comparisons was performed for all of the above analyses; results are considered significant at a threshold of p < 0.05 (FDR corrected). Between-group differences in clinical variables were calculated using analysis of variance (ANOVA) with Bonferroni post-hoc correction, apart from categorical data, for which Pearson’s Chi Squared test was used.

RESULTS

Clinical characteristics

AD patients performed worse compared with healthy controls and patients with MCI on the CDR-SB (p < 0.001), MMSE (p < 0.001), MoCA (p < 0.001), and RAVLT-immediate (p < 0.001, Table 1). MCI patients performed worse on the CDR-SB (p < 0.001), MoCA (p < 0.001), and RAVLT-immediate tests (p < 0.001) compared with healthy controls. 88%of AD patients failed to remember any words during the RAVLT-forgetting test, limiting the utility of this test tests for discriminating between patients with AD. One outlying CDR-SB score was identified in the AD group; this patient was subsequently removed from all correlation analyses involving CDR-SB.

There were significantly more females in the control group compared to patients with MCI (p = 0.009) and AD (p = 0.007). Gender was therefore used as a covariate, alongside age, in all analyses. One or more APOE ɛ4 alleles was present in 30%of healthy controls, 40%of MCI, and 56%of AD patients. No significant difference was detected between groups; however, APOE status was not available in one healthy control, two patients with MCI, and eight patients with AD (Table 1). There were significantly more amyloid positive patients in the AD group (87%) compared to MCI (41%, p = 0.001), and healthy controls (25%, p < 0.001), but no significant difference between healthy controls and MCI.

Alterations in microstructural connectivity in MCI and AD

After thresholding, 1,566 edges remained for analysis. Reduced connectivity was detected between multiple brain areas in AD compared to controls, predominantly involving medial temporal structures, in particular the hippocampus (Fig. 1, Table 2). Many of the regions showing reduced connectivity in AD, also showed reduced connectivity in MCI compared to controls; however, they did not remain significant after FDR correction. Increased connectivity was also noted, between the right amygdala and right temporal structures in AD compared to controls.

Fig. 1

Widespread connectome changes in Alzheimer’s disease. Connectogram rendering of whole brain structural connectivity in Alzheimer’s disease (AD) compared to healthy controls. [¹⁸F]AV-1451 SUVr and [¹⁸F]AV-45 SUVr (above controls mean), and strength (percent of controls mean) of each region are indicated. Connecting lines between regions represent significantly decreased (red) or increased (green) connectivity (FDR corrected). Line weights are proportional to differences in connectivity (change in mean number of streamlines compared to healthy controls).

Table 2

Microstructural connectivity in mild cognitive impairment and Alzheimer’s disease compared to healthy controls

Connection between		Mean connections (SD)						p value
Region 1	Region 2	HC		MCI		AD		MCI versus HC	AD versus HC	AD versus MCI
L. hippocampus	L. inferior temporal	4,149	(1,500)	2,963	(1,785)	1,843	(1,184)	0.038^*	0.005	0.005^*
L. hippocampus	L. thalamus	8,823	(3,469)	6,228	(3,027)	4,448	(2,675)	0.009^*	0.006	0.028^*
L. hippocampus	L. supramarginal	315	(126)	240	(166)	133	(87)	0.99	0.009	0.003^*
L. hippocampus	L. middle temporal	2,355	(876)	1,612	(1,095)	1,164	(774)	0.017^*	0.019	0.37
L. hippocampus	L. posterior cingulate	196	(84)	168	(88)	99	(53)	0.99	0.019	0.001^*
L. hippocampus	L. postcentral	322	(201)	234	(187)	112	(123)	0.99	0.019	0.010^*
L. hippocampus	L. inferior parietal	687	(281)	517	(343)	332	(160)	0.99	0.021	0.011^*
L. hippocampus	L. precentral	768	(628)	514	(448)	235	(286)	0.99	0.029	0.029^*
L. hippocampus	L. putamen	2,103	(1,063)	1,490	(1,123)	935	(727)	0.99	0.029	0.031^*
L. hippocampus	L. fusiform	12,018	(5,022)	8,939	(3,720)	7,027	(3,944)	0.033^*	0.029	0.39
L. hippocampus	L. paracentral	153	(123)	109	(87)	53	(56)	0.99	0.029	0.024^*
L. hippocampus	L. caudate	961	(746)	761	(606)	325	(316)	0.99	0.031	0.007^*
L. hippocampus	L. superior frontal	389	(336)	279	(248)	118	(134)	0.99	0.039	0.020^*
L. hippocampus	R. thalamus	149	(83)	106	(81)	68	(57)	0.99	0.042	0.34
L. inferior temporal	L. thalamus	1,186	(382)	1,089	(307)	823	(261)	0.99	0.019	< 0.001^*
L. inferior temporal	L. lateral orbitofrontal	215	(128)	149	(103)	84	(67)	0.99	0.025	0.019^*
L. superior temporal	L. lateral orbitofrontal	321	(148)	265	(138)	188	(94)	0.99	0.035	0.031^*
L. pars opercularis	L. putamen	4,217	(1,276)	4,018	(923)	3,171	(923)	0.99	0.038	0.002^*
L. rostral middle frontal	L. caudate	8,684	(2,050)	7,294	(2,300)	6,410	(2,034)	0.036^*	0.039	0.44
R. hippocampus	R. inferior temporal	5,325	(1,474)	4,160	(2,028)	2,846	(1,617)	0.046^*	0.005	0.003^*
R. hippocampus	R. posterior cingulate	336	(110)	271	(126)	200	(82)	0.036^*	0.009	0.012^*
R. hippocampus	R. thalamus	8,221	(2,502)	5,504	(2,562)	5,005	(2,373)	< 0.001^*	0.015	0.70
R. hippocampus	R. caudate	1,324	(596)	939	(649)	589	(504)	0.047^*	0.019	0.030^*
R. hippocampus	R. supramarginal	391	(169)	270	(193)	191	(112)	0.029^*	0.025	0.37
R. hippocampus	R. putamen	2,796	(1,609)	1,881	(1,183)	1,367	(992)	0.017^*	0.025	0.43
R. hippocampus	R. superior frontal	639	(360)	408	(325)	261	(272)	0.025^*	0.029	0.40
R. hippocampus	R. postcentral	514	(353)	327	(259)	210	(143)	0.032^*	0.029	0.40
R. hippocampus	R. amygdala	2,279	(822)	1,722	(879)	1,467	(672)	0.013^*	0.029	0.51
R. hippocampus	R. precentral	1,312	(897)	832	(602)	549	(456)	0.034^*	0.037	0.44
R. hippocampus	R. lateral orbitofrontal	228	(185)	106	(78)	98	(91)	0.001^*	0.044	0.91
R. inferior temporal	R. putamen	2,310	(621)	1,980	(720)	1,491	(759)	0.99	0.029	0.006^*
R. inferior parietal	L. postcentral	116	(69)	105	(60)	61	(25)	0.99	0.044	0.004^*
R. precuneus	L. postcentral	358	(189)	353	(186)	193	(71)	1.0	0.044	< 0.001^*
R. isthmus cingulate	L. precuneus	1,586	(537)	1,555	(481)	1,197	(413)	0.99	0.047	0.004^*
R. precuneus	L. precuneus	2,983	(1,052)	2,981	(1,070)	1,942	(1,006)	1.0	0.047	< 0.001^*
R. entorhinal	R. superior temporal	418	(318)	406	(204)	753	(502)	1.0	0.029	< 0.001^*
R. amygdala	R. superior temporal	1,880	(830)	2,354	(990)	3,318	(1,172)	0.99	0.005	< 0.001^*
R. amygdala	R. banks STS	55	(24)	67	(30)	94	(47)	0.99	0.019	0.002^*

AD, Alzheimer’s disease; Banks STS, Banks of the superior temporal sulcus; HC, healthy controls; L, Left; MCI, mild cognitive impairment; R, right; SD, standard deviation. ^*Uncorrected p value provided where FDR corrected p > 0.05. All other p values are FDR corrected.

Reduced hippocampal connectivity in MCI and AD

When assessing hippocampal connectivity to ipsilateral limbic structures, after thresholding, 15 edges remained for analysis, in line with areas of maximal hippocampal connectivity in the literature [98]. There were widespread reductions in connectivity in AD compared to controls (Fig. 2B, Supplementary Table 1). Notably, significant reductions were also present in the MCI cohort compared to controls; between the hippocampus and ipsilateral thalamus, bilaterally, and the right hippocampus and ipsilateral orbitofrontal cortex and amygdala (Fig. 2A, Supplementary Table 1).

Fig. 2

Reduced hippocampal connectivity in mild cognitive impairment and Alzheimer’s disease. When examining connectivity between the hippocampus and ipsilateral limbic structures, reduced connectivity was detected in mild cognitive impairment (MCI; A) and Alzheimer’s disease (AD; B) compared to healthy controls. [¹⁸F]AV-1451 SUVr and [¹⁸F]AV-45 SUVr (above controls mean), and strength (percent of controls mean) of each region are indicated. Connecting lines between regions represent significantly reduced connectivity (FDR corrected). Line weights are proportional to differences in connectivity (change in mean number of streamlines compared to healthy controls).

Global cortical [¹⁸F]AV-45 SUVr is not related to hippocampal [¹⁸F]AV-45 SUVr or hippocampal strength

There was no significant relationship between cortical [¹⁸F]AV-45 SUVr in a composite region comprising the frontal, cingulate, parietal, and temporal regions bilaterally and either hippocampal [¹⁸F]AV-1451 SUVr or hippocampal strength in healthy controls, MCI, or AD.

Reduced hippocampal connectivity is related to [¹⁸F]AV-45 SUVr in the target region of those same hippocampal connections

Reductions in connectivity between the hippocampus and ipsilateral brain regions correlated with increased [¹⁸F]AV-45 SUVr in the target regions of those same connections in MCI (r = –0.33, p = 0.003) and AD (r = –0.31, p = 0.006) compared to healthy controls, and AD compared to MCI (r = –0.25, p =0.023). There were no such relationships for [¹⁸F]AV-1451 SUVr.

Reduced regional strength in MCI and AD

In AD, there was reduced strength in the left (p = 0.006) and right (p = 0.022) hippocampus, left accumbens (p = 0.049), right caudate (p = 0.049), and right putamen (p = 0.049) compared with healthy controls (Supplementary Table 2). MCI patients had reduced strength in the left (p = 0.013 uncorrected) and right (p = 0.002 uncorrected) hippocampus, however this did not remain significant after FDR correction for multiple comparisons.

Region of interest [¹⁸F]AV-1451 and [¹⁸F]AV-45 PET findings in MCI and AD

[¹⁸F]AV-1451 PET

Of those 64 ROIs assessed, [¹⁸F]AV-1451 SUVr was increased in 61 ROIs in AD compared with healthy controls including both entorhinal cortices (p < 0.001), amygdalae (p < 0.001), and hippocampi (p < 0.001), and 48 ROIs in AD compared with MCI including the left entorhinal cortex (p = 0.012), left amygdala (p = 0.035), right amygdala (p = 0.015), left hippocampus (p = 0.003), and right hippocampus (p = 0.016, Supplementary Table 3). Prior to correction for multiple comparisons, ten of these same regions had increased [¹⁸F]AV-1451 SUVr in MCI compared with healthy controls including the left entorhinal cortex (p = 0.007), right entorhinal cortex (p = 0.029), left amygdala (p = 0.002), right amygdala (p = 0.005), left hippocampus (p = 0.003), and right hippocampus (p = 0.008); however, they did not remain significant after FDR correction.

Prior to partial volume correction increased choroid plexus [¹⁸F]AV-1451 SUVr correlated with higher hippocampal [¹⁸F]AV-1451 SUVr (r = 0.29, p = 0.006). After partial volume correction there was no longer any relationship between choroid plexus and hippocampal [¹⁸F]AV-1451 SUVr (r = 0.01, p = 0.94).

[¹⁸F]AV-45 PET

AV-45 SUVr was increased in AD compared to controls in 69 of 82 ROIs examined including in the nucleus accumbens bilaterally (p < 0.001). In 53 ROIs [¹⁸F]AV-45 SUVr was increased in AD compared with MCI including in the left (p = 0.020) and right nucleus accumbens (p < 0.018; Supplementary Table 4). While [¹⁸F]AV-45 SUVr was almost universally increased in AD compared with healthy controls and MCI, there was no significant difference in [¹⁸F]AV-45 SUVr within the entorhinal cortices or right hippocampus between groups, while in the left hippocampus, [¹⁸F]AV-45 SUVr was reduced in AD (p = 0.035) compared with healthy controls. There was no significant increase in [¹⁸F]AV-45 SUVr in MCI compared to healthy controls in any region.

Global correlations between [¹⁸F]AV-1451 SUVr, [¹⁸F]AV-45 SUVr, and strength in AD and MCI

Increased global [¹⁸F]AV-1451 SUVr correlated with reduced strength in MCI (r = –0.07, p = 0.001) and AD (r = –0.10, p < 0.001), while there was no correlation between [¹⁸F]AV-45 SUVr and strength in controls, MCI, or AD (Supplementary Table 5). Increased global [¹⁸F]AV-1451 SUVr correlated with increased [¹⁸F]AV-45 SUVr in controls (r = 0.15, p < 0.001), MCI (r = 0.18, p < 0.001), and AD (r = 0.16, p < 0.001, Supplementary Table 5).

Regional correlations between [¹⁸F]AV-1451 SUVr, [¹⁸F]AV-45 SUVr, and strength in AD and MCI

In the hippocampus, increased [¹⁸F]AV-1451 SUVr correlated with reduced hippocampal strength in MCI (r = –0.55, p = 0.049, Fig. 3A) and AD (r =–0.56, p = 0.046; Fig. 3B). Additionally, increased [¹⁸F]AV-1451 SUVr correlated with reduced strength in the middle temporal gyrus (r = –0.69, p = 0.006, Fig. 3C) and inferior parietal gyrus (r = –0.56, p =0.046, Fig. 3D) in AD. There was no significant correlation between [¹⁸F]AV-45 SUVr and strength in any region after correction for multiple comparisons in AD or MCI.

Fig. 3

Correlations between connectome strength and [¹⁸F]AV-1451 SUVr in mild cognitive impairment and Alzheimer’s disease. Increased [¹⁸F]AV-1451 standardized uptake value ratio (SUVr) correlated with reduced strength within the hippocampi in both mild cognitive impairment (MCI; A) and Alzheimer’s disease (AD; B). In AD, increased [18F]AV-1451 SUVr also correlated with reduced strength in the middle temporal gyri (C) and inferior parietal gyri (D).

Correlations between regional [¹⁸F]AV-1451 and [¹⁸F]AV-45 SUVr in MCI

In MCI patients, increased [¹⁸F]AV-1451 SUVr correlated with increased [¹⁸F]AV-45 SUVr across a number of neocortical areas within Thal’s stage 1 of amyloid-β deposition in AD, with the exception of the isthmus of the cingulate gyrus which is within Thal stage 2 (Supplementary Table 6). No significant correlations between [¹⁸F]AV-1451 SUVr and [¹⁸F]AV-45 SUVr were present in AD.

Correlations with cognitive scores in MCI and AD

Connectome strength and cognitive scores

In AD, lower RAVLT-immediate scores correlated with reduced left hippocampal strength (r = 0.69, p = 0.024, Fig. 4A). Strength of the right hippocampus also correlated with the RAVLT-immediate score; however, this did not survive FDR correction for multiple comparisons (r = 0.56, p = 0.005 uncorrected). Furthermore, lower MoCA scores correlated with reduced strength of the right banks of the superior temporal sulcus (r = 0.67, p = 0.038) in AD (Fig. 4B).

Fig. 4

Correlations between connectome strength and cognitive scores in mild cognitive impairment and Alzheimer’s disease. In Alzheimer’s disease (AD), lower Rey Auditory Verbal Learning Test (RAVLT)-immediate scores correlated with reduced left hippocampal strength (A), and lower Montreal Cognitive Assessment (MoCA) scores correlated with reduced strength in the right banks of the superior temporal sulcus (STS) (B). Lower scores on both of these cognitive scales represent increased impairment. In mild cognitive impairment (MCI), reduced strength in the left inferior temporal gyrus correlated with reduced Functional Activities Questionnaire (FAQ) scores (C), indicating increased impairment, while reduced right temporal pole strength in MCI correlated with higher RAVLT-forgetting scores, indicating greater impairment (D).

In MCI, higher FAQ scores correlated with increased strength in the left inferior temporal gyrus (r = 0.61, p = 0.031, Fig. 4C), while higher RAVLT-forgetting scores correlated with reduced strength of the right temporal pole (r = –0.60, p = 0.035, Fig. 4D), indicating that MCI patients with increased strength in the right temporal pole forgot fewer words.

[¹⁸F]AV-1451 and [¹⁸F]AV-45 PET and cognitive scores

Increased impairment on CDR-SB and MMSE scores correlated with higher [¹⁸F]AV-1451 SUVr in regions corresponding to Braak stages I-IV in MCI, predominantly within limbic structures (Fig. 5, Supplementary Table 7). In the AD cohort, higher CDR-SB scores correlated with higher [¹⁸F]AV-1451 SUVr across regions corresponding to Braak stages III-VI, with the notable exception of regions in Braak stage I-II in which correlations were present in MCI (Fig. 5, Supplementary Table 8). There were no significant correlations between [¹⁸F]AV-1451 SUVr and MMSE in patients with AD. This pattern was noted after completion of our analysis; we have not statistically compared these grouped results between patient groups due to the inherent bias in such an analysis at this stage.

Fig. 5

Significant correlations between cognitive scores and [¹⁸F]AV-1451 SUVr in regions corresponding to Braak’s staging in mild cognitive impairment and Alzheimer’s disease. Graphical representation of differing distributions of correlation of cognitive scores with [¹⁸F]AV-1451 SUVr in mild cognitive impairment (MCI) and Alzheimer’s disease (AD). In MCI, increased impairment on MMSE and CDR-SB correlated with [¹⁸F]AV-1451 SUVr in regions corresponding to early Braak stages (I-IV), while the CDR-SB in AD correlated with [¹⁸F]AV-1451 SUVr in regions corresponding to the later Braak stages (III-VI).

There was no significant correlation between [¹⁸F]AV-45 SUVr in any region with CDR-SB or MMSE scores in AD or MCI.

The effect of APOE status on the connectome, [¹⁸F]AV-1451 and [¹⁸F]AV-45 SUVr

Carriers of one or more copies of APOE ɛ4 had a significantly higher global [¹⁸F]AV-45 SUVr (1.48±0.09) than non-carriers in AD (1.35±0.16; p = 0.045). APOE ɛ4 allele carrier status had no significant effect on global [¹⁸F]AV-1451 SUVr or strength in MCI or AD, or global [¹⁸F]AV-45 SUVr in MCI. There was no interaction between APOE4 status and regional [¹⁸F]AV-1451 SUVr, [¹⁸F]AV-45 SUVr, or strength in any group.

DISCUSSION

Using multi-modal imaging, we demonstrate microstructural connectivity changes between regions of increased tau pathology in MCI and AD, shedding light on the relationship between the accumulation of tau pathology and the neurodegenerative process of AD. Reductions in hippocampal connectivity predominate in the AD connectome, with more marginal changes in MCI. Reduced hippocampal connectivity was associated with hippocampal tau accumulation in both MCI and AD, while cognitive scores reflect the sequential spread of tau pathology. Our findings suggest that the accumulation of tau pathology underlies the reductions in microstructural connectivity that contribute to the development of cognitive decline in MCI and AD.

Reductions in hippocampal connectivity are associated with tau pathology in MCI and AD

When all brain regions are examined together, we demonstrate that regions with higher tau deposition, in general, have reduced structural connectivity in both MCI and AD. However, this negative correlation between tau deposition and regional strength is weaker than would be expected were tau deposition the only determinant of degeneration. Many areas with a high tau burden have minimal reductions in connectivity strength. A far stronger relationship between tau deposition and strength is present when assessed within specific regions. Hippocampal tau deposition is strongly associated with reduced hippocampal connectivity in both MCI and AD; suggesting that tau is central to the degenerative process in the hippocampus, acting prior to the clinical onset of AD.

Reductions in hippocampal connectivity are related to amyloid-β pathology in the target of those connections in MCI and AD

We have shown that reductions in hippocampal connectivity in MCI and AD are significantly related to the levels of amyloid-β within the targets of those same hippocampal connections. Hippocampal connections to areas with greater amyloid-β deposition had greater reductions in connectivity to those same areas in both MCI and AD when compared to healthy controls, and in AD compared to MCI. We could not demonstrate a relationship between hippocampal strength and global amyloid-β deposition, potentially reflecting the fact that examining amyloid-β deposition across the whole brain overlooks the local effects of a sequential disease process.

The distribution of amyloid-β in AD, distant from the hippocampal epicenter of the neurodegenerative process, has long been one of the unresolved problems of the amyloid hypothesis. While progressive amyloid-β deposition has been related to disease progression, the nature of this relationship remains unclear [2 , 69]. Here we show that amyloid-β deposition distant to the hippocampus is directly related to reductions in hippocampal connectivity to these same areas in MCI and AD; suggesting that amyloid-β is directly, and spatially linked to the neurodegenerative process. Given that there is no relationship between local amyloid-β deposition and reductions in microstructural connectivity, it appears unlikely that this effect is a direct toxic effect of amyloid-β upon the neurons that make up these connections. Instead, such distant effects may fit with proposed transneuronal, prion-like, or downstream effects of amyloid-β in the neurodegenerative process, as well as theories of a synergistic role for amyloid-β and tau [49 , 99–102]. Both amyloid-β and tau appear to have an effect upon the hippocampus, tau acting locally, and amyloid-β via its hippocampal connections.

Degeneration of hippocampal connections underlies cognitive decline in MCI and AD

Increasing evidence implicates degeneration in the Papez circuit, and in particular the limbic thalamus, in the early stages of AD [90 , 104]. Degeneration of these pathways, integral to the consolidation of memory and emotional processing, provide an anatomical correlate for the clinical presentation of AD. We demonstrate bilateral reductions in hippocampo-thalamic connectivity to be foremost among the microstructural aberrations of MCI, alongside reduced right hippocampal connectivity to the ipsilateral amygdala and lateral orbitofrontal gyrus. Reduced hippocampal connectivity was far more widespread in AD, involving connections to numerous limbic structures, as well as the dorsal striatum, parietal and temporal areas. These reductions are not limited to hippocampal connections, with significantly reduced connectivity between numerous other limbic structures and components of the Papez circuit. Alongside evidence from lesioning and pharmacological studies suggesting a role the striatum in learning and memory, and morphological studies demonstrating associations between striatal atrophy and cognitive decline in AD [105 –109], our results provide microstructural evidence for a loss hippocampo-striatal connectivity as part of the disease process of AD.

Using a number of cognitive measures, we demonstrate that reductions of microstructural connectivity are associated with cognitive decline in both MCI and AD. There are strong correlations between reduced strength of the left inferior temporal gyrus and FAQ scores in MCI; and reduced strength in the right banks STS and MoCA scores in AD. Cognitive assessments which target specific cognitive domains are warranted to assess the effects of degeneration within specific localized brain regions. The RAVLT score, a more straightforward test of episodic verbal memory, appears better suited in this regard. Impaired performance on RAVLT is associated specifically with lesions to medial and anterior temporal structures [110 –115].

We demonstrate that the strength of the left hippocampus strongly correlates with RAVLT-immediate scores in AD, while in MCI the strength of the right temporal pole is strongly correlated with RAVLT-forgetting scores. Dysfunction in the anterior temporal lobes is increasingly implicated in disorders of semantic processing while functional activity of the anterior temporal network, which includes the temporal pole, is associated with performance in declarative memory tasks [116, 117]. In this context the association between RAVLT-forgetting score and right temporal pole strength suggests that early disruption in the anterior temporal network may account for some of the cognitive deficit in MCI. The use of the different components of the RAVLT herein was necessitated by the failure of the majority of the AD cohort to provide any correct answers on the more challenging RAVLT-forgetting test.

Previous studies have used machine learning approaches to predict RAVLT-immediate scores with moderate accuracy from atrophy on MRI using multiple brain areas in a regression model [53]. Our results suggest that regional connectivity may be superior to these volumetric measures and warrant further examination in longitudinal studies.

Cognitive decline reflects the accumulation of tau pathology in MCI and AD

The pattern of tau accumulation in AD is well established by histological studies [17, 66]. Here, we demonstrate that the entorhinal and hippocampal accumulation of tau is strongly correlated with the extent of cognitive decline in MCI, prior to the clinical onset of AD. In the AD cohort, in whom there is already significant tau deposition within the early Braak regions, the progression of cognitive decline instead follows the deposition of tau within the limbic and neocortical regions characterized by the later Braak stages (III-VI).

It has already been shown that baseline levels of tau and baseline cognitive scores can predict the rate of cognitive decline in AD [118, 119]. This study goes further by demonstrating, in vivo, that cognitive decline across the spectrum of MCI and AD follows the pathological progression of tau pathology. The correlation between tau deposition and hippocampal strength in both MCI and AD further supports a direct role of tau in the pathological process, providing a microstructural correlate for the cognitive decline.

In contrast to the relationships between tau, connectivity, and cognitive scores described here, no such relationships were present between amyloid-β deposition and either the cognitive or microstructural changes of AD. We demonstrate significantly increased amyloid-β deposition throughout the brain in AD, yet these increases are not related to global or regional connectome measures, or to cognition in AD or MCI. There was a weak but significant global relationship between levels of amyloid-β and tau deposition present in HC, MCI, and AD. However, given the divergent patterns of tau and amyloid-β accumulation in aging and AD, it is unclear whether these correlations represent causation or are simply the sequelae of two overlapping processes [66, 119].

In a condition characterized by amyloid-β deposition and hippocampal degeneration, it is striking that the hippocampus itself has the lowest [¹⁸F]AV-45 SUVr among all brain regions studied, with sequentially less hippocampal amyloid-β detected in patients with MCI and AD, reaching significance in the left hippocampus of patients with AD compared to controls. Histological studies have noted the paucity of amyloid-β within the hippocampus, limited primarily to the CA1 region in the middle and latter stages of the disease [66, 67]. The decreased hippocampal amyloid-β reported here may be partially artifactual, resultant on increased amyloid-β deposition in the cerebellum, to which the PET is normalized. Regardless, the hippocampi are clearly relatively devoid of amyloid-β when compared to the rest of the brain.

Limitations

The relatively small number of patients in this study was resultant on the requirement for adequate imaging across multiple imaging modalities. This will have limited the ability to detect subtle changes, as will the necessary but stringent statistical correction for multiple comparisons, where many regions are examined simultaneously. While some atlases provide smaller ROIs than those provided by the Desikan-Killiany atlas and subcortical segmentation used herein, demonstrating significant effects would have been challenging given the statistical correction for multiple comparisons required. More focused examination of the results reported here, in future studies, will allow confirmation of our findings and examination of the differential effects of the neurodegenerative process within smaller regions and tracts, while longitudinal studies can further examine any causal relationships suggested by this pseudo-longitudinal study.

Off target binding of the [¹⁸F]AV-1451 tracer within the basal ganglia and thalamus necessarily meant that these regions were excluded from this study. While a large number of studies have reported specific binding of the [¹⁸F]AV-1451 tracer when examining hippocampal tau deposition, including autoradiographic studies, some have reported erroneous results within the hippocampi due to the spill-off effects of off-target binding from the adjacent choroid plexus, despite partial volume correction [120 –123]. The partial volume correction used herein appears to be resilient to these effects; while the potential for some contamination from the choroid plexus remains, the significant relationship between choroid plexus and hippocampal [¹⁸F]AV-1451 SUVr is no longer present after partial volume correction, while significant relationships between hippocampal [¹⁸F]AV-1451 SUVr and other metrics reported here would regardless appear extremely unlikely to be driven by binding from the choroid plexus.

This study does not explore the duration of pathological tau exposure in any region; those areas with recent tau deposition may be expected to have less degeneration despite high levels of tau than those regions with longer exposure. However, given increasing evidence of selective vulnerability of different neuronal subtypes, it is likely that underlying vulnerability of certain neuronal populations is a significant determinant of degeneration [31 , 124–126]. Disruptions of hippocampal neurogenesis [127], neurotrophic signaling [128], and endosomal transport [129], have all been implicated in the selective vulnerability of particular neuronal populations in AD, as have the type and proportion of different neuronal subtypes [128, 130], yet the exact cause remains to be established.

In conclusion, we demonstrate widespread disruption of limbic circuitry in AD, alongside similar, if more marginal changes to hippocampal connections in MCI. While tau pathology is related to local microstructural changes and cognitive scores, reductions in hippocampal connectivity are correlated with levels of amyloid-β in the target regions of those connections. Amyloid-β and tau appear to have differential effects upon the hippocampus; tau acting locally, and amyloid-β via its hippocampal connections.

Footnotes

ACKNOWLEDGMENTS

Data collection and sharing for this project was funded by the Alzheimer’s Disease Neuroimaging Initiative (ADNI) (National Institutes of Health Grant U01 AG024904) and DOD ADNI (Department of Defense award number W81XWH-12-2-0012). ADNI is funded by the National Institute on Aging, the National Institute of Biomedical Imaging and Bioengineering, and through generous contributions from the following: AbbVie, Alzheimer’s Association; Alzheimer’s Drug Discovery Foundation; Araclon Biotech; BioClinica, Inc.; Biogen; Bristol-Myers Squibb Company; CereSpir, Inc.; Cogstate; Eisai Inc.; Elan Pharmaceuticals, Inc.; Eli Lilly and Company; EuroImmun; F. Hoffmann-La Roche Ltd and its affiliated company Genentech, Inc.; Fujirebio; GE Healthcare; IXICO Ltd.; Janssen Alzheimer Immunotherapy Research & Development, LLC.; Johnson & Johnson Pharmaceutical Research & Development LLC.; Lumosity; Lundbeck; Merck & Co., Inc.; Meso Scale Diagnostics, LLC.; NeuroRx Research; Neurotrack Technologies; Novartis Pharmaceuticals Corporation; Pfizer Inc.; Piramal Imaging; Servier; Takeda Pharmaceutical Company; and Transition Therapeutics. The Canadian Institutes of Health Research is providing funds to support ADNI clinical sites in Canada. Private sector contributions are facilitated by the Foundation for the National Institutes of Health (). The grantee organization is the Northern California Institute for Research and Education, and the study is coordinated by the Alzheimer’s Therapeutic Research Institute at the University of Southern California. ADNI data are disseminated by the Laboratory for Neuro Imaging at the University of Southern California.

Professor Marios Politis research is supported by Michael J Fox Foundation for Parkinson’s Research, Edmond and Lilly Safra Foundation, CHDI Foundation, Glaxo Wellcome R&D, Life Molecular Imaging, Invicro, Curium, Medical Research Council (UK), AVID radiopharmaceuticals, National Institute for Health Research, Alzheimer’s Research UK, and European Commission IMI2 fund. Doctor Josh King-Robson’s research is funded by the National Institute for Health Research (NIHR). The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Braak

, Alafuzoff

, Arzberger

, Kretzschmar

, Del Tredici

(2006) Staging of Alzheimer disease-associated neurofibrillary pathology using paraffin sections and immunocytochemistry. Acta Neuropathol 112, 389–404.

Kametani

, Hasegawa

(2018) Reconsideration of amyloid hypothesis and tau hypothesis in Alzheimer’s disease. Front Neurosci 12, 25.

Bateman

, Xiong

, Benzinger

, Fagan

, Goate

, Fox

, Marcus

, Cairns

, Xie

, Blazey

, Holtzman

, Santacruz

, Buckles

, Oliver

, Moulder

, Aisen

, Ghetti

, Klunk

, McDade

, Martins

, Masters

, Mayeux

, Ringman

, Rossor

, Schofield

, Sperling

, Salloway

, Morris

, Dominantly Inherited Alzheimer Network (2012) Clinical and biomarker changes in dominantly inherited Alzheimer’s disease. N Engl J Med 367, 795–804.

Hardy

, Higgins

(1992) Alzheimer’s disease: The amyloid cascade hypothesis. Science 256, 184–185.

Kang

, Lemaire

, Unterbeck

, Salbaum

, Masters

, Grzeschik

, Multhaup

, Beyreuther

, Muller-Hill

(1987) The precursor of Alzheimer’s disease amyloid A4 protein resembles a cell-surface receptor. Nature 325, 733–736.

Goate

, Chartier-Harlin

, Mullan

, Brown

, Crawford

, Fidani

, Giuffra

, Haynes

, Irving

, James

, Mant

, Newton

, Rooke

, Roques

, Talbot

, Pericak-Vance

, Roses

, Williamson

, Rossor

, Owen

, Hardy

(1991) Segregation of a missense mutation in the amyloid precursor protein gene with familial Alzheimer’s disease. Nature 349, 704–706.

Hussain

, Powell

, Howlett

, Tew

, Meek

, Chapman

, Gloger

, Murphy

, Southan

, Ryan

, Smith

, Simmons

, Walsh

, Dingwall

, Christie

(1999) Identification of a novel aspartic protease (Asp 2) as beta-secretase. Mol Cell Neurosci 14, 419–427.

Sherrington

, Rogaev

, Liang

, Rogaeva

, Levesque

, Ikeda

, Chi

, Lin

, Li

, Holman

, Tsuda

, Mar

, Foncin

, Bruni

, Montesi

, Sorbi

, Rainero

, Pinessi

, Nee

, Chumakov

, Pollen

, Brookes

, Sanseau

, Polinsky

, Wasco

, Da Silva

, Haines

, Perkicak-Vance

, Tanzi

, Roses

, Fraser

, Rommens

, St George-Hyslop

(1995) Cloning of a gene bearing missense mutations in early-onset familial Alzheimer’s disease. Nature 375, 754–760.

Levy-Lahad

, Wasco

, Poorkaj

, Romano

, Oshima

, Pettingell

, Yu

, Jondro

, Schmidt

, Wang

, et al. (1995) Candidate gene for the chromosome 1 familial Alzheimer’s disease locus. Science 269, 973–977.

10.

Rogaev

, Sherrington

, Rogaeva

, Levesque

, Ikeda

, Liang

, Chi

, Lin

, Holman

, Tsuda

, Mar

, Sorbi

, Nacmias

, Piacentini

, Amaducci

, Chumakov

, Cohen

, Lannfelt

, Fraser

, Rommens

, St George-Hyslop

(1995) Familial Alzheimer’s disease in kindreds with missense mutations in a gene on chromosome 1 related to the Alzheimer’s disease type 3 gene. Nature 376, 775–778.

11.

Karran

, Mercken

, De Strooper

(2011) The amyloid cascade hypothesis for Alzheimer’s disease: An appraisal for the development of therapeutics. Nat Rev Drug Discov 10, 698–712.

12.

Chiti

, Dobson

(2017) Protein misfolding, amyloid formation, and human disease: A summary of progress over the last decade. Annu Rev Biochem 86, 27–68.

13.

Hardy

, De Strooper

(2017) Alzheimer’s disease: Where next for anti-amyloid therapies? Brain 140, 853–855.

14.

Aizenstein

, Nebes

, Saxton

, Price

, Mathis

, Tsopelas

, Ziolko

, James

, Snitz

, Houck

, Bi

, Cohen

, Lopresti

, DeKosky

, Halligan

, Klunk

(2008) Frequent amyloid deposition without significant cognitive impairment among the elderly. Arch Neurol 65, 1509–1517.

15.

Katzman

, Terry

, DeTeresa

, Brown

, Davies

, Fuld

, Renbing

, Peck

(1988) Clinical, pathological, and neurochemical changes in dementia: A subgroup with preserved mental status and numerous neocortical plaques. Ann Neurol 23, 138–144.

16.

Panza

, Lozupone

, Logroscino

, Imbimbo

(2019) A critical appraisal of amyloid-beta-targeting therapies for Alzheimer disease. Nat Rev Neurol 15, 73–88.

17.

Braak

, Braak

(1995) Staging of Alzheimer’s disease-related neurofibrillary changes.271-278; discussion. Neurobiol Aging 16, 278–284.

18.

Nelson

, Jicha

, Schmitt

, Liu

, Davis

, Mendiondo

, Abner

, Markesbery

(2007) Clinicopathologic correlations in a large Alzheimer disease center autopsy cohort: Neuritic plaques and neurofibrillary tangles “do count” when staging disease severity. J Neuropathol Exp Neurol 66, 1136–1146.

19.

Pickett

, Henstridge

, Allison

, Pitstick

, Pooler

, Wegmann

, Carlson

, Hyman

, Spires-Jones

(2017) Spread of tau down neural circuits precedes synapse and neuronal loss in the rTgTauEC mouse model of early Alzheimer’s disease. Synapse 71, e21965.

20.

Nakamura

, Greenwood

, Binder

, Bigio

, Denial

, Nicholson

, Zhou

, Lu

(2012) Proline isomer-specific antibodies reveal the early pathogenic tau conformation in Alzheimer’s disease. Cell 149, 232–244.

21.

Kondo

, Shahpasand

, Mannix

, Qiu

, Moncaster

, Chen

, Yao

, Lin

, Driver

, Sun

, Wei

, Luo

, Albayram

, Huang

, Rotenberg

, Ryo

, Goldstein

, Pascual-Leone

, McKee

, Meehan

, Zhou

, Lu

(2015) Antibody against early driver of neurodegeneration cis P-tau blocks brain injury and tauopathy. Nature 523, 431–436.

22.

, Gillberg

, Bergfors

, Marutle

, Nordberg

(2013) Amyloid tracers detect multiple binding sites in Alzheimer’s disease brain tissue. Brain 136, 2217–2227.

23.

Mishra

, Gordon

, Su

, Christensen

, Friedrichsen

, Jackson

, Hornbeck

, Balota

, Cairns

, Morris

, Ances

, Benzinger

TLS

(2017) AV-1451 PET imaging of tau pathology in preclinical Alzheimer disease: Defining a summary measure. Neuroimage 161, 171–178.

24.

Chien

, Bahri

, Szardenings

, Walsh

, Mu

, Su

, Shankle

, Elizarov

, Kolb

(2013) Early clinical PET imaging results with the novel PHF-tau radioligand [F-18]-T807. J Alzheimers Dis 34, 457–468.

25.

Smith

, Puschmann

, Scholl

, Ohlsson

, van Swieten

, Honer

, Englund

, Hansson

(2016) 18F-AV-1451 tau PET imaging correlates strongly with tau neuropathology in MAPT mutation carriers. Brain 139, 2372–2379.

26.

Schultz

, Gordon

, Mishra

, Su

, Perrin

, Cairns

, Morris

, Ances

, Benzinger

TLS

(2018) Widespread distribution of tauopathy in preclinical Alzheimer’s disease. Neurobiol Aging 72, 177–185.

27.

Johnson

, Sperling

, Gidicsin

, Carmasin

, Maye

, Coleman

, Reiman

, Sabbagh

, Sadowsky

, Fleisher

, Murali Doraiswamy

, Carpenter

, Clark

, Joshi

, Lu

, Grundman

, Mintun

, Pontecorvo

, Skovronsky

, AV45-A11 study group (2013) Florbetapir (F18-AV-45) PET to assess amyloid burden in Alzheimer’s disease dementia, mild cognitive impairment, and normal aging. Alzheimers Dement 9, S72–83.

28.

Schwarz

, Yu

, Miller

, Shcherbinin

, Dickson

, Navitsky

, Joshi

, Devous

Sr. , Mintun

(2016) Regional profiles of the candidate tau PET ligand 18F-AV-1451 recapitulate key features of Braak histopathological stages. Brain 139, 1539–1550.

29.

Marquie

, Normandin

, Meltzer

, Siao Tick Chong

, Andrea

, Anton-Fernandez

, Klunk

, Mathis

, Ikonomovic

, Debnath

, Bien

, Vanderburg

, Costantino

, Makaretz

, DeVos

, Oakley

, Gomperts

, Growdon

, Domoto-Reilly

, Lucente

, Dickerson

, Frosch

, Hyman

, Johnson

, Gomez-Isla

(2017) Pathological correlations of [F-18]-AV-1451 imaging in non-Alzheimer tauopathies. Ann Neurol 81, 117–128.

30.

Schöll

, Schonhaut

, Lockhart

, Vogel

, Baker

, Schwimmer

, Ossenkoppele

, Rabinovici

, Jagust

(2015) In vivo braak staging using 18F-AV1451 Tau PET imaging. Alzheimers Dement 11, P4.

31.

Marquie

, Normandin

, Vanderburg

, Costantino

, Bien

, Rycyna

, Klunk

, Mathis

, Ikonomovic

, Debnath

, Vasdev

, Dickerson

, Gomperts

, Growdon

, Johnson

, Frosch

, Hyman

, Gomez-Isla

(2015) Validating novel tau positron emission tomography tracer [F-18]-AV-1451 (T807) on postmortem brain tissue. Ann Neurol 78, 787–800.

32.

Choi

, Schneider

, Bennett

, Beach

, Bedell

, Zehntner

, Krautkramer

, Kung

, Skovronsky

, Hefti

, Clark

(2012) Correlation of amyloid PET ligand florbetapir F 18 binding with Abeta aggregation and neuritic plaque deposition in postmortem brain tissue. Alzheimer Dis Assoc Disord 26, 8–16.

33.

Schöll

, Lockhart

, Schonhaut

, O’Neil

, Janabi

, Ossenkoppele

, Baker

, Vogel

, Faria

, Schwimmer

, Rabinovici

, Jagust

(2016) PET imaging of tau deposition in the aging human brain. Neuron 89, 971–982.

34.

Mitchell

, Shiri-Feshki

(2009) Rate of progression of mild cognitive impairment to dementia–meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand 119, 252–265.

35.

Tucholka

, Grau-Rivera

, Falcon

, Rami

, Sanchez-Valle

, Llado

, Gispert

, Molinuevo

, Alzheimer’s Disease Neuroimaging Initiative (2018) Structural connectivity alterations along the alzheimer’s disease continuum: Reproducibility across two independent samples and correlation with cerebrospinal fluid amyloid-beta and tau. J Alzheimers Dis 61, 1575–1587.

36.

, Mori

, Chan

, Ma

(2019) Connectome-wide network analysis of white matter connectivity in Alzheimer’s disease. Neuroimage Clin 22, 101690.

37.

, Wang

, Chou

, Wang

, He

, Lin

(2010) Diffusion tensor tractography reveals abnormal topological organization in structural cortical networks in Alzheimer’s disease. J Neurosci 30, 16876–16885.

38.

Prescott

, Guidon

, Doraiswamy

, Roy Choudhury

, Liu

, Petrella

, Alzheimer’s Disease Neuroimaging Initiative (2014) The Alzheimer structural connectome: Changes in cortical network topology with increased amyloid plaque burden. Radiology 273, 175–184.

39.

Filippi

, Basaia

, Canu

, Imperiale

, Magnani

, Falautano

, Comi

, Falini

, Agosta

(2020) Changes in functional and structural brain connectome along the Alzheimer’s disease continuum. Mol Psychiatry 25, 230–239.

40.

Pereira

(2020) Detecting early changes in Alzheimer’s disease with graph theory. Brain Commun 2, fcaa129.

41.

Vermunt

, Dicks

, Wang

, Dincer

, Flores

, Keefe

, Berman

, Cash

, Chhatwal

, Cruchaga

, Fox

, Ghetti

, Graff-Radford

, Hassenstab

, Karch

, Laske

, Levin

, Masters

, McDade

, Mori

, Morris

, Noble

, Perrin

, Schofield

, Xiong

, Scheltens

, Visser

, Bateman

, Benzinger

TLS

, Tijms

, Gordon

(2020) Single-subject grey matter network trajectories over the disease course of autosomal dominant Alzheimer’s disease.fcaa. Brain Commun 2, 102.

42.

David

, Eberle

, Delev

, Gaubatz

, Prillwitz

, Wagner

, Schoene-Bake

J-C

, Luechters

, Radbruch

, Wabbels

, Schramm

, Weber

, Surges

, Elger

, Rüber

(2021) Multi-scale image analysis and prediction of visual field defects after selective amygdalohippocampectomy. Sci Rep 11, 1444.

43.

Smith

, Tournier

, Calamante

, Connelly

(2015) SIFT2: Enabling dense quantitative assessment of brain white matter connectivity using streamlines tractography. Neuroimage 119, 338–351.

44.

Smith

, Tournier

, Calamante

, Connelly

(2012) Anatomically-constrained tractography: Improved diffusion MRI streamlines tractography through effective use of anatomical information. Neuroimage 62, 1924–1938.

45.

Wen

, Risacher

, Xie

, Li

, Harezlak

, Farlow

, Unverzagt

, Gao

, Apostolova

, Saykin

, Wu

(2021) Tau-related white-matter alterations along spatially selective pathways. Neuroimage 226, 117560.

46.

Strain

, Smith

, Beaumont

, Roe

, Gordon

, Mishra

, Adeyemo

, Christensen

, Su

, Morris

, Benzinger

TLS

, Ances

(2018) Loss of white matter integrity reflects tau accumulation in Alzheimer disease defined regions. Neurology 91, e313–e318.

47.

Kantarci

, Murray

, Schwarz

, Reid

, Przybelski

, Lesnick

, Zuk

, Raman

, Senjem

, Gunter

, Boeve

, Knopman

, Parisi

, Petersen

, Jack

, Dickson

(2017) White-matter integrity on DTI and the pathologic staging of Alzheimer’s disease. Neurobiol Aging 56, 172–179.

48.

Sintini

, Schwarz

, Martin

, Graff-Radford

, Machulda

, Senjem

, Reid

, Spychalla

, Drubach

, Lowe

, Jack

, Josephs

, Whitwell

(2019) Regional multimodal relationships between tau, hypometabolism, atrophy, and fractional anisotropy in atypical Alzheimer’s disease. Hum Brain Mapp 40, 1618–1631.

49.

Jacobs

HIL

, Hedden

, Schultz

, Sepulcre

, Perea

, Amariglio

, Papp

, Rentz

, Sperling

, Johnson

(2018) Structural tract alterations predict downstream tau accumulation in amyloid-positive older individuals. Nat Neurosci 21, 424–431.

50.

Iaccarino

, Tammewar

, Ayakta

, Baker

, Bejanin

, Boxer

, Gorno-Tempini

, Janabi

, Kramer

, Lazaris

, Lockhart

, Miller

, O’Neil

, Ossenkoppele

, Rosen

, Schonhaut

, Jagust

, Rabinovici

(2018) Local and distant relationships between amyloid, tau and neurodegeneration in Alzheimer’s disease. Neuroimage Clin 17, 452–464.

51.

Ossenkoppele

, Schonhaut

, Baker

, O’Neil

, Janabi

, Ghosh

, Santos

, Miller

, Bettcher

, Gorno-Tempini

, Miller

, Jagust

, Rabinovici

(2015) Tau, amyloid, and hypometabolism in a patient with posterior cortical atrophy. Ann Neurol 77, 338–342.

52.

O’Bryant

, Waring

, Cullum

, Hall

, Lacritz

, Massman

, Lupo

, Reisch

, Doody

, Texas Alzheimer’s Research Consortium (2008) Staging dementia using Clinical Dementia Rating Scale Sum of Boxes scores: A Texas Alzheimer’s research consortium study. Arch Neurol 65, 1091–1095.

53.

Moradi

, Hallikainen

, Hanninen

, Tohka

, Alzheimer’s Disease Neuroimaging Initiative (2017) Rey’s Auditory Verbal Learning Test scores can be predicted from whole brain MRI in Alzheimer’s disease. Neuroimage Clin 13, 415–427.

54.

Estevez-Gonzalez

, Kulisevsky

, Boltes

, Otermin

, Garcia-Sanchez

(2003) Rey verbal learning test is a useful tool for differential diagnosis in the preclinical phase of Alzheimer’s disease: Comparison with mild cognitive impairment and normal aging. Int J Geriatr Psychiatry 18, 1021–1028.

55.

Schmidt

(1996) Rey auditory verbal learning test: A handbook, Western Psychological Services Los Angeles, CA.

56.

Pfeffer

, Kurosaki

, Harrah

, Chance

, Filos

(1982) Measurement of functional activities in older adults in the community. J Gerontol 37, 323–329.

57.

Teng

, Becker

, Woo

, Knopman

, Cummings

, Lu

(2010) Utility of the Functional Activities Questionnaire for distinguishing mild cognitive impairment from very mild Alzheimer disease. Alzheimer Dis Assoc Disord 24, 348–353.

58.

Liu

C-C

, Kanekiyo

, Xu

, Bu

(2013) Apolipoprotein E and Alzheimer disease: Risk, mechanisms and therapy. Nat Rev Neurol 9, 106–118.

59.

Zavaliangos-Petropulu

, Nir

, Thomopoulos

, Reid

, Bernstein

, Borowski

, Jack

Jr. , Weiner

, Jahanshad

, Thompson

(2019) Diffusion MRI indices and their relation to cognitive impairment in brain aging: The updated multi-protocol approach in ADNI3. Front Neuroinform 13, 2.

60.

Jagust

, Bandy

, Chen

, Foster

, Landau

, Mathis

, Price

, Reiman

, Skovronsky

, Koeppe

, Alzheimer’s Disease Neuroimaging Initiative (2010) The Alzheimer’s Disease Neuroimaging Initiative positron emission tomography core. Alzheimers Dement 6, 221–229.

61.

Jagust

, Landau

, Koeppe

, Reiman

, Chen

, Mathis

, Price

, Foster

, Wang

(2015) The Alzheimer’s Disease Neuroimaging Initiative 2 PET Core: 2015. Alzheimers Dement 11, 757–771.

62.

Tournier

, Smith

, Raffelt

, Tabbara

, Dhollander

, Pietsch

, Christiaens

, Jeurissen

, Yeh

, Connelly

(2019) MRtrix3: A fast, flexible and open software framework for medical image processing and visualisation. Neuroimage 202, 116137.

63.

Jenkinson

, Smith

(2001) Global optimisation method for robust affine registration of brain images. Med Image Anal 5, 143–156.

64.

Jenkinson

, Bannister

, Brady

, Smith

(2002) Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage 17, 825–841.

65.

Fischl

(2012) FreeSurfer. Neuroimage 62, 774–781.

66.

Braak

, Braak

(1991) Neuropathological stageing of Alzheimer-related changes. Acta Neuropathol 82, 239–259.

67.

Braak

, Braak

(1997) Frequency of stages of Alzheimer-related lesions in different age categories. Neurobiol Aging 18, 351–357.

68.

Serrano-Pozo

, Frosch

, Masliah

, Hyman

(2011) Neuropathological alterations in Alzheimer disease. Cold Spring Harb Perspect Med 1, a006189–a006189.

69.

Thal

, Rüb

, Orantes

, Braak

(2002) Phases of Aβ-deposition in the human brain and its relevance for the development of AD. Neurology 58, 1791–1800.

70.

Rousset

, Ma

, Evans

(1998) Correction for partial volume effects in PET: Principle and validation. J Nucl Med 39, 904–911.

71.

Baker

, Maass

, Jagust

(2017) Considerations and code for partial volume correcting [(18)F]-AV-1451 tau PET data. Data Brief 15, 648–657.

72.

Maass

, Landau

, Baker

, Horng

, Lockhart

, La Joie

, Rabinovici

, Jagust

, Alzheimer’s Disease Neuroimaging Initiative (2017) Comparison of multiple tau-PET measures as biomarkers in aging and Alzheimer’s disease. Neuroimage 157, 448–463.

73.

Diedrichsen

, Maderwald

, Kuper

, Thurling

, Rabe

, Gizewski

, Ladd

, Timmann

(2011) Imaging the deep cerebellar nuclei: A probabilistic atlas and normalization procedure. Neuroimage 54, 1786–1794.

74.

Penny

, Friston

, Ashburner

, Kiebel

S, E.

, Nichols

(2007) Statistical Parametric Mapping.

75.

Joshi

, Pontecorvo

, Clark

, Carpenter

, Jennings

, Sadowsky

, Adler

, Kovnat

, Seibyl

, Arora

, Saha

, Burns

, Lowrey

, Mintun

, Skovronsky

, Florbetapir

18 Study Investigators (2012) Performance characteristics of amyloid PET with florbetapir F 18 in patients with alzheimer’s disease and cognitively normal subjects. J Nucl Med 53, 378–384.

76.

Marquie

, Verwer

, Meltzer

, Kim

SJW

, Aguero

, Gonzalez

, Makaretz

, Siao Tick Chong

, Ramanan

, Amaral

, Normandin

, Vanderburg

, Gomperts

, Johnson

, Frosch

, Gomez-Isla

(2017) Lessons learned about [F-18]-AV-1451 off-target binding from an autopsy-confirmed Parkinson’s case. Acta Neuropathol Commun 5, 75.

77.

Guo

, Korman

, La Joie

, Shaw

, Trojanowski

, Jagust

, Landau

(2020) Normalization of CSF pTau measurement by Aβ40 improves its performance as a biomarker of Alzheimer’s disease. Alzheimers Res Ther 12, 97.

78.

Landau

, Murphy

, Ward

, Jagust

(2015) Florbetapir (AV45) processing methods. Alzheimer’s Disease Neuroimaging Initiative, https://adni.bitbucket.io/reference/docs/UCBERKELEYAV45/ADNI_AV45_MethodsJagustLab_06.25.15.pdf.

79.

Landau

, Koeppe

, Jagust

(2011) Florbetaben processing and positivity threshold derivation. Alzheimer’s Disease Neuro-imaging Initiative, https://adni.bitbucket.io/reference/docs/UCBERKELEYFBB/UCBerkeley_FBB_Methods_04.11.19.pdf.

80.

Veraart

, Novikov

, Christiaens

, Ades-Aron

, Sijbers

, Fieremans

(2016) Denoising of diffusion MRI using random matrix theory. Neuroimage 142, 394–406.

81.

Andersson

JLR

, Sotiropoulos

(2016) An integrated approach to correction for off-resonance effects and subject movement in diffusion MR imaging. Neuroimage 125, 1063–1078.

82.

Veraart

, Fieremans

, Novikov

(2016) Diffusion MRI noise mapping using random matrix theory. Magn Reson Med 76, 1582–1593.

83.

Smith

, Jenkinson

, Woolrich

, Beckmann

, Behrens

, Johansen-Berg

, Bannister

, De Luca

, Drobnjak

, Flitney

, Niazy

, Saunders

, Vickers

, Zhang

, De Stefano

, Brady

, Matthews

(2004) Advances in functional and structural MR image analysis and implementation as FSL. Neuroimage 23Suppl 1, S208–219.

84.

Tournier

, Calamante

, Connelly

(2007) Robust determination of the fibre orientation distribution in diffusion MRI: Non-negativity constrained super-resolved spherical deconvolution. Neuroimage 35, 1459–1472.

85.

Tournier

, Calamante

, Gadian

, Connelly

(2004) Direct estimation of the fiber orientation density function from diffusion-weighted MRI data using spherical deconvolution. Neuroimage 23, 1176–1185.

86.

Smith

, Tournier

, Calamante

, Connelly

(2015) The effects of SIFT on the reproducibility and biological accuracy of the structural connectome. Neuroimage 104, 253–265.

87.

Krzywinski

, Schein

, Birol

, Connors

, Gascoyne

, Horsman

, Jones

, Marra

(2009) Circos: An information aesthetic for comparative genomics. Genome Res 19, 1639–1645.

88.

Rubinov

, Sporns

(2010) Complex network measures of brain connectivity: Uses and interpretations. Neuroimage 52, 1059–1069.

89.

Papez

(1995) A proposed mechanism of emotion. 1937. J Neuropsychiatry Clin Neurosci 7, 103–112.

90.

Aggleton

, Pralus

, Nelson

, Hornberger

(2016) Thalamic pathology and memory loss in early Alzheimer’s disease: Moving the focus from the medial temporal lobe to Papez circuit. Brain 139, 1877–1890.

91.

Rolls

(2015) Limbic systems for emotion and for memory, but no single limbic system. Cortex 62, 119–157.

92.

Nakano

(1998) The limbic system: An outline and brief history of its concept. Neuropathology 18, 211–214.

93.

Landau

, Breault

, Joshi

, Pontecorvo

, Mathis

, Jagust

, Mintun

(2013) Amyloid-β imaging with Pittsburgh Compound B and florbetapir: Comparing radiotracers and quantification methods. J Nucl Med 54, 70–77.

94.

Nichols

, Hayasaka

(2003) Controlling the familywise error rate in functional neuroimaging: A comparative review. Stat Methods Med Res 12, 419–446.

95.

Genovese

, Lazar

, Nichols

(2002) Thresholding of statistical maps in functional neuroimaging using the false discovery rate. Neuroimage 15, 870–878.

96.

Benjamini

, Hochberg

(1995) Controlling The False Discovery Rate - a practical and powerful approach to multiple testing. J R Stat Soc Series B 57, 289–300.

97.

Motulsky

, Brown

(2006) Detecting outliers when fitting data with nonlinear regression - a new method based on robust nonlinear regression and the false discovery rate. BMC Bioinformatics 7, 123.

98.

Maller

, Welton

, Middione

, Callaghan

, Rosenfeld

, Grieve

(2019) Revealing the hippocampal connectome through super-resolution 1150-direction diffusion MRI. Sci Rep 9, 2418.

99.

Aguzzi

, Rajendran

(2009) The transcellular spread of cytosolic amyloids, prions, and prionoids. Neuron 64, 783–790.

100.

Walker

, Diamond

, Duff

, Hyman

(2013) Mechanisms of protein seeding in neurodegenerative diseases. JAMA Neurol 70, 304.

101.

Nussbaum

, Schilling

, Cynis

, Silva

, Swanson

, Wangsanut

, Tayler

, Wiltgen

, Hatami

, Rönicke

, Reymann

, Hutter-Paier

, Alexandru

, Jagla

, Graubner

, Glabe

, Demuth

, Bloom

(2012) Prion-like behaviour and tau-dependent cytotoxicity of pyroglutamylated amyloid-β. Nature 485, 651–655.

102.

Spires-Jones

, Hyman

(2014) The intersection of amyloid beta and tau at synapses in Alzheimer’s disease. Neuron 82, 756–771.

103.

Zarei

, Patenaude

, Damoiseaux

, Morgese

, Smith

, Matthews

, Barkhof

, Rombouts

, Sanz-Arigita

, Jenkinson

(2010) Combining shape and connectivity analysis: An MRI study of thalamic degeneration in Alzheimer’s disease. Neuroimage 49, 1–8.

104.

Bishnoi

, Surya

, McCall

(2017) Structural integrity of white matter tracts in papez circuit during the course of Alzheimer’s disease. Alzheimers Dement 13, P1379.

105.

Goodroe

, Starnes

, Brown

(2018) The complex nature of hippocampal-striatal interactions in spatial navigation. Front Hum Neurosci 12, 250.

106.

de Jong

, Ferrarini

, van der Grond

, Milles

, Reiber

, Westendorp

, Bollen

, Middelkoop

, van Buchem

(2011) Shape abnormalities of the striatum in Alzheimer’s disease. J Alzheimers Dis 23, 49–59.

107.

Pievani

, Bocchetta

, Boccardi

, Cavedo

, Bonetti

, Thompson

, Frisoni

(2013) Striatal morphology in early-onset and late-onset Alzheimer’s disease: A preliminary study. Neurobiol Aging 34, 1728–1739.

108.

Cho

, Kim

, Ye

, Kim

, Yoon

, Noh

, Kim

, Kang

, Chin

, Kim

, Lee

, Na

, Seong

, Seo

(2014) Shape changes of the basal ganglia and thalamus in Alzheimer’s disease: A three-year longitudinal study. J Alzheimers Dis 40, 285–295.

109.

Packard

, Knowlton

(2002) Learning and memory functions of the basal ganglia. Annu Rev Neurosci 25, 563–593.

110.

Schoenberg

, Clifton

, Sever

, Vale

(2018) Neuropsychology outcomes following trephine epilepsy surgery: The inferior temporal gyrus approach for amygdalohippocampectomy in medically refractory mesial temporal lobe epilepsy. Neurosurgery 82, 833–841.

111.

Biesbroek

, Van Zandvoort

MJE

, Kappelle

, Schoo

, Kuijf

, Velthuis

, Biessels

, Postma

(2015) Distinct anatomical correlates of discriminability and criterion setting in verbal recognition memory revealed by lesion-symptom mapping. Hum Brain Mapp 36, 1292–1303.

112.

Sudo

, De Souza

, Drummond

, Assuncao

, Teldeschi

, Oliveira

, Rodrigues

, Santiago-Bravo

, Calil

, Lima

, Erthal

, Bernardes

, Monteiro

, Tovar-Moll

, Mattos

(2019) Inter-method and anatomical correlates of episodic memory tests in the Alzheimer’s Disease spectrum. PLoS One 14, e0223731.

113.

Wolk

, Dickerson

(2011) Fractionating verbal episodic memory in Alzheimer’s disease. Neuroimage 54, 1530–1539.

114.

Nakamura

, Kubota

(1996) The primate temporal pole: Its putative role in object recognition and memory. Behav Brain Res 77, 53–77.

115.

Majdan

, Sziklas

, Jones-Gotman

(1996) Performance of healthy subjects and patients with resection from the anterior temporal lobe on matched tests of verbal and visuoperceptual learning. J Clin Exp Neuropsychol 18, 416–430.

116.

Gour

, Ranjeva

, Ceccaldi

, Confort-Gouny

, Barbeau

, Soulier

, Guye

, Didic

, Felician

(2011) Basal functional connectivity within the anterior temporal network is associated with performance on declarative memory tasks. Neuroimage 58, 687–697.

117.

Visser

, Jefferies

, Lambon Ralph

(2010) Semantic processing in the anterior temporal lobes: A meta-analysis of the functional neuroimaging literature. J Cogn Neurosci 22, 1083–1094.

118.

Kennedy

, Cutter

, Wang

, Schneider

(2015) Using baseline cognitive severity for enriching Alzheimer’s disease clinical trials: How does Mini-Mental State Examination predict rate of change? Alzheimers Dement (N Y) 1, 46–52.

119.

Pontecorvo

, Devous

, Kennedy

, Navitsky

, Lu

, Galante

, Salloway

, Doraiswamy

, Southekal

, Arora

, McGeehan

, Lim

, Xiong

, Truocchio

, Joshi

, Shcherbinin

, Teske

, Fleisher

, Mintun

(2019) A multicentre longitudinal study of flortaucipir (18F) in normal ageing, mild cognitive impairment and Alzheimer’s disease dementia. Brain 142, 1723–1735.

120.

Lee

, Jacobs

HIL

, Marquié

, Becker

, Andrea

, Jin

, Schultz

, Frosch

, Gómez-Isla

, Sperling

, Johnson

(2018) 18F-flortaucipir binding in choroid plexus: Related to race and hippocampus signal. J Alzheimers Dis 62, 1691–1702.

121.

Wang

, Benzinger

, Su

, Christensen

, Friedrichsen

, Aldea

, McConathy

, Cairns

, Fagan

, Morris

, Ances

(2016) Evaluation of tau imaging in staging Alzheimer disease and revealing interactions between β-amyloid and tauopathy. JAMA Neurol 73, 1070.

122.

Lowe

, Curran

, Fang

, Liesinger

, Josephs

, Parisi

, Kantarci

, Boeve

, Pandey

, Bruinsma

, Knopman

, Jones

, Petrucelli

, Cook

, Graff-Radford

, Dickson

, Petersen

, Jack

, Murray

(2016) An autoradiographic evaluation of AV-1451 Tau PET in dementia. Acta Neuropathol Commun 4, 58.

123.

Passamonti

, Vázquez Rodríguez

, Hong

, Allinson

, Williamson

, Borchert

, Sami

, Cope

, Bevan-Jones

, Jones

, Arnold

, Surendranathan

, Mak

, Su

, Fryer

, Aigbirhio

, O’Brien

, Rowe

(2017) 18F-AV-1451 positron emission tomography in Alzheimer’s disease and progressive supranuclear palsy. Brain 140, 781–791.

124.

, Hardy

, Duff

(2018) Selective vulnerability in neurodegenerative diseases. Nature Neurosci 21, 1350–1358.

125.

Götz

, Schonrock

, Vissel

, Ittner

(2009) Alzheimer’s disease selective vulnerability and modeling in transgenic mice. J Alzheimers Dis 18, 243–251.

126.

Mattsson

, Schott

, Hardy

, Turner

, Zetterberg

(2016) Selective vulnerability in neurodegeneration: Insights from clinical variants of Alzheimer’s disease. J Neurol Neurosurg Psychiatry 87, 1000–1004.

127.

Kirschen

, Kéry

, Ge

(2018) The hippocampal neuro-glio-vascular network: Metabolic vulnerability and potential neurogenic regeneration in disease. Brain Plasticity 3, 129–144.

128.

Ginsberg

, Malek-Ahmadi

, Alldred

, Che

, Elarova

, Chen

, Jeanneteau

, Kranz

, Chao

, Counts

, Mufson

(2019) Selective decline of neurotrophin and neurotrophin receptor genes within CA1 pyramidal neurons and hippocampus proper: Correlation with cognitive performance and neuropathology in mild cognitive impairment and Alzheimer’s disease. Hippocampus 29, 422–439.

129.

Small

(2014) Isolating pathogenic mechanisms embedded within the hippocampal circuit through regional vulnerability. Neuron 84, 32–39.

130.

Muratore

, Zhou

, Liao

, Fernandez

, Taylor

, Lagomarsino

, Pearse

, 2nd, Rice

, Negri

, He

, Srikanth

, Callahan

, Shin

, Zhou

, Bennett

, Noggle

, Love

, Selkoe

, Young-Pearse

(2017) Cell-type dependent Alzheimer’s disease phenotypes: Probing the biology of selective neuronal vulnerability. Stem Cell Rep 9, 1868–1884.

Associations Between Amyloid and Tau Pathology,and Connectome Alterations,in Alzheimer’s Disease and Mild Cognitive Impairment

Abstract

Background:

Objective:

Methods:

Results:

Conclusion:

Keywords

INTRODUCTION

MATERIALS AND METHODS

Participants

Clinical assessments

Scanning procedures

Magnetic resonance imaging

Positron emission tomography

Imaging data analysis

Structural MRI analysis and regions of interest

PET data analysis

Connectome construction

Regional and global correlations

Statistical analysis

RESULTS

Clinical characteristics

Alterations in microstructural connectivity in MCI and AD

Reduced hippocampal connectivity in MCI and AD

Global cortical [18F]AV-45 SUVr is not related to hippocampal [18F]AV-45 SUVr or hippocampal strength

Reduced hippocampal connectivity is related to [18F]AV-45 SUVr in the target region of those same hippocampal connections

Reduced regional strength in MCI and AD

Region of interest [18F]AV-1451 and [18F]AV-45 PET findings in MCI and AD

[18F]AV-1451 PET

[18F]AV-45 PET

Global correlations between [18F]AV-1451 SUVr, [18F]AV-45 SUVr, and strength in AD and MCI

Regional correlations between [18F]AV-1451 SUVr, [18F]AV-45 SUVr, and strength in AD and MCI

Correlations between regional [18F]AV-1451 and [18F]AV-45 SUVr in MCI

Correlations with cognitive scores in MCI and AD

Connectome strength and cognitive scores

[18F]AV-1451 and [18F]AV-45 PET and cognitive scores

The effect of APOE status on the connectome, [18F]AV-1451 and [18F]AV-45 SUVr

DISCUSSION

Reductions in hippocampal connectivity are associated with tau pathology in MCI and AD

Reductions in hippocampal connectivity are related to amyloid-β pathology in the target of those connections in MCI and AD

Degeneration of hippocampal connections underlies cognitive decline in MCI and AD

Cognitive decline reflects the accumulation of tau pathology in MCI and AD

Limitations

Footnotes

ACKNOWLEDGMENTS

References

Global cortical [¹⁸F]AV-45 SUVr is not related to hippocampal [¹⁸F]AV-45 SUVr or hippocampal strength

Reduced hippocampal connectivity is related to [¹⁸F]AV-45 SUVr in the target region of those same hippocampal connections

Region of interest [¹⁸F]AV-1451 and [¹⁸F]AV-45 PET findings in MCI and AD

[¹⁸F]AV-1451 PET

[¹⁸F]AV-45 PET

Global correlations between [¹⁸F]AV-1451 SUVr, [¹⁸F]AV-45 SUVr, and strength in AD and MCI

Regional correlations between [¹⁸F]AV-1451 SUVr, [¹⁸F]AV-45 SUVr, and strength in AD and MCI

Correlations between regional [¹⁸F]AV-1451 and [¹⁸F]AV-45 SUVr in MCI

[¹⁸F]AV-1451 and [¹⁸F]AV-45 PET and cognitive scores

The effect of APOE status on the connectome, [¹⁸F]AV-1451 and [¹⁸F]AV-45 SUVr